Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →A股可比公司分析/行业估值对标。当用户说"可比公司"、"估值对标"、"行业估值"、"comps"、"comparable company"、"同行对比"、"XX跟同行比怎么样"、"XX板块估值"、"可比公司分析"、"comps analysis"时触发。MUST USE when user asks about comparable company analysis, peer valuation comparison, or industry valuation benchmarking (comps). 通过 cn-stock-data 获取标的及可比公司的财务指标和实时行情，构建多维度估值对标表。输出 Excel（公式驱动）或 Markdown 对比表。支持投行估值表风格（formal）和快速对标表风格（brief）。
.claude/skills/aifinlab-a-share-comps/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-10 | ✗→✓ | ▲ Improved | -3% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 11% | 0% |
| case-06 | ✓→✗ | ▼ Worse | -56% | 0% |
| case-14 | ✓→✗ | ▼ Worse | -68% | 0% |
| case-09 | ✓→✗ | ▼ Worse | -53% | 0% |
bashSCRIPTS="$SKILLS_ROOT/cn-stock-data/scripts" # 各公司财务指标 python "$SCRIPTS/cn_stock_data.py" finance --code [CODE] # 各公司实时行情（市值、PE、PB 等） python "$SCRIPTS/cn_stock_data.py" quote --code [CODE1],[CODE2],[CODE3] # 各公司近 1 年日线（用于股价走势对比） python "$SCRIPTS/cn_stock_data.py" kline --code [CODE] --freq daily --start [1年前日期]
补充：通过 web 搜索确认行业分类、可比公司池选择的合理性。
可比公司选择注意事项：
对标的和每家可比公司获取：
finance — 最新财务指标（ROE、毛利率、净利率、增速等）quote — 实时行情（市值、PE、PB）参见 references/valuation-metrics.md 的指标定义和行业处理规则。
核心估值列（必含）： | 代码 | 名称 | 市值(亿) | PE(TTM) | PB | ROE(%) | 营收增速(%) | 净利润增速(%) | 毛利率(%) | 净利率(%) |
进阶列（formal 模式）： | EV/EBITDA | PS | PEG | 资产负债率(%) | 经营现金流/利润 | 股息率(%) |
计算可比公司组的：
Excel 输出规则（formal 模式）：
Markdown 输出（brief 模式）：
| 维度 | formal（投行估值表） | brief（快速对标表） | |------|-------------------|--------------------| | 输出格式 | Excel (.xlsx) | Markdown 表格 | | 指标列数 | 12-15 列 | 6-8 列 | | 统计行 | 均值+中位数+最大+最小 | 仅中位数 | | 图表 | PE/PB 散点图数据 | 无 | | 结论 | 客观描述估值位置 | 可加个人判断 | | 免责声明 | 需要 | 不需要 |
=市值单元格/净利润单元格| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 36,731 | 10,584 | -71% | 1 | 1 | 0% | 8,289 | 2,006 | -76% | 0 | 0 | — |
case-02 | fail→fail | 23,796 | 12,546 | -47% | 1 | 1 | 0% | 4,270 | 1,687 | -60% | 0 | 0 | — |
case-03 | fail→fail | 11,591 | 8,480 | -27% | 1 | 1 | 0% | 1,759 | 1,645 | -6% | 0 | 0 | — |
case-04 | fail→fail | 19,352 | 16,320 | -16% | 1 | 1 | 0% | 3,111 | 2,456 | -21% | 0 | 0 | — |
case-05 | fail→fail | 26,701 | 13,420 | -50% | 1 | 1 | 0% | 3,974 | 2,002 | -50% | 0 | 0 | — |
case-06 | pass→fail | 27,277 | 11,544 | -58% | 1 | 1 | 0% | 3,696 | 1,643 | -56% | 0 | 0 | — |
case-14 | pass→fail | 29,090 | 10,767 | -63% | 1 | 1 | 0% | 4,705 | 1,490 | -68% | 0 | 0 | — |
case-07 | pass→pass | 20,981 | 33,872 | +61% | 1 | 1 | 0% | 3,587 | 6,629 | +85% | 0 | 0 | — |
case-08 | pass→pass | 18,906 | 19,235 | +2% | 1 | 1 | 0% | 2,914 | 4,032 | +38% | 0 | 0 | — |
case-09 | pass→fail | 23,888 | 36,972 | +55% | 1 | 1 | 0% | 3,324 | 1,546 | -53% | 0 | 0 | — |
case-10 | fail→pass | 18,496 | 45,402 | +145% | 1 | 1 | 0% | 2,638 | 2,547 | -3% | 0 | 0 | — |
case-11 | fail→pass | 25,478 | 23,640 | -7% | 1 | 1 | 0% | 4,121 | 4,582 | +11% | 0 | 0 | — |
case-12 | pass→fail | 17,944 | 55,768 | +211% | 1 | 1 | 0% | 2,538 | 1,913 | -25% | 0 | 0 | — |
case-13 | fail→fail | 14,113 | 7,746 | -45% | 1 | 1 | 0% | 2,053 | 1,557 | -24% | 0 | 0 | — |
case-15 | pass→fail | 24,493 | 6,765 | -72% | 1 | 1 | 0% | 4,011 | 1,483 | -63% | 0 | 0 | — |
case-16 | pass→fail | 23,037 | 7,194 | -69% | 1 | 1 | 0% | 3,669 | 1,437 | -61% | 0 | 0 | — |
case-17 | pass→pass | 25,738 | 18,465 | -28% | 1 | 1 | 0% | 3,731 | 4,133 | +11% | 0 | 0 | — |
case-18 | pass→pass | 19,554 | 25,487 | +30% | 1 | 1 | 0% | 2,876 | 4,840 | +68% | 0 | 0 | — |
case-19 | fail→fail | 26,675 | 10,628 | -60% | 1 | 1 | 0% | 3,826 | 1,972 | -48% | 0 | 0 | — |
case-20 | fail→fail | 14,239 | 5,507 | -61% | 1 | 1 | 0% | 2,309 | 1,993 | -14% | 0 | 0 | — |
case-21 | pass→pass | 16,020 | 15,029 | -6% | 1 | 1 | 0% | 2,327 | 3,078 | +32% | 0 | 0 | — |
case-22 | pass→pass | 20,166 | 16,098 | -20% | 1 | 1 | 0% | 3,176 | 3,827 | +20% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted, and 9 counted toward the lift figure. The other 13 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of -18 percentage points is the difference between those two pass rates over the 9 comparable cases. 13 cases got worse with the skill loaded, and they are included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.